|InterJournal Genetics, 598
|Manuscript Number: |
Submission Date: 20628
|Interdisciplinary Conceptual Model Blending|
Subject(s): BG.00, CX.13, CX.3, CX.41, CX.65, CX.11
It is becoming increasingly common that research efforts, and the development of systems to support them, are being undertaken by multidisciplinary teams. Life science multidisciplinary research has become the norm, rather than an innovation. There are several reasons for this trend. Insights from several points of view provide a richer understanding of issues and more opportunities for solutions. In addition, the thought pattern from one discipline may provide new insights to another discipline. In many disciplines the research in part(s) of the domain has reached the stage where exploring issues and advances in adjoining parts, and in the interaction of parts, is warranted. In hierarchical systems, especially living systems, research at individual levels is different in kind from research at others, and integration across levels has become possible and desirable. Multidisciplinary research presents a marvelous opportunity, but also creates serious problems. Merging of the disciplines' conceptualizations must occur, at least in the (separate) minds of the collaborators. This merging must leverage the expertise of individual discipline members as well as that of general-purpose designers. Systems that support this type of research are complex systems, with significant semantic mismatch problems: at the user's perceptual interface, the system's conceptual model and the system's (and database's) design model levels. The information portrayed to scientists through the system's perceptual interface is structures in accordance with its own natural model, but must be portrayed in accordance with a model natural to its user. Various parts of these systems exhibit different conceptual organizations, but users need to envision a unified model. Finally, the increasing use of computer databases for organizing disparate research results in data integration problems. Models for each database or data source are designed independently, in accordance with a domain's conceptual model. These models are further specialized to a particular research effort, then encoded using general-purpose data models. The independence of development and the differing cultures of the fields cause incompatibilities between models and programming interfaces. The notation's general-purpose nature loses (filters) insights and intuition from domains' natural illustrations and explanations of key models. This paper explores a mechanism and an analytical methodology for integration of multiple disciplines' models. It is based on distilling the inherent structure of each model, blending models to create the structure for the integrated domain and creating views of this blended structure for each participating discipline. The approach has four aspects. The underlying structure of the "natural" models is extracted by analysis of the metaphorical underpinnings of those models.Natural graphic depictions and explanations are integrated with general-purpose models. Models are blended using the character of one to underlie semantics taken from others. Finally, a framework for visualization and understanding of the blended domain is created using the natural depictions, explanations and underlying metaphors. This technique provides a framework for understanding, organizing and supporting interdisciplinary work. It improves the conceptual modeling process by explicitly integrating more domain intuition and insight into the process. We will illustrate the mechanisms and a methodology for use with examples from interdisciplinary projects in molecular biology and ecology.
|Submit referee report/comment|